A combined Codex and Claude Code plugin marketplace for Outcome Engineering — the product engineering paradigm where a durable map of your product, maintained as a Product Tree, serves as the authoritative source of truth for all implementation.
This repository publishes two plugin surfaces from the same source tree:
.claude-pluginpackages for Claude Code skills and thin agents.codex-pluginpackages for Codex skill bundles
AGENTS.md and CLAUDE.md are separate root instruction files that share the product-owned repository guidance and carry harness-specific managed Product Tree sections.
/bootstrapinterviews you about your product, then scaffolds a Product Tree — the durable map that drives all implementation.
- RTFM: Follow state-of-the-art (SOTA) model prompting guidance, such as structured prompts based on XML tags
- KILO: Keep It Local and Observable — the golden source for all specifications lives locally within the product's Git repository
- ABC: Always Be Converging — the Product Tree is the desired state; all activities are driven by it, not the other way around
- DCI: Deterministic Context Injection — the Product Tree constructs a deterministic context injection plan for the model
npm install -g @outcomeeng/spxThe spx CLI is the developer tool for Product Tree maintenance and validation. Required by all engineering plugins.
claude plugin marketplace add outcomeeng/pluginscodex plugin marketplace add outcomeeng/pluginsCodex registers the marketplace source in the user's ~/.codex/config.toml and reads the shared .codex-plugin bundles from it.
# Product Tree methodology (requires spx CLI)
claude plugin install product-tree@outcomeeng
# Language plugins (install per product, require spx CLI)
claude plugin install python@outcomeeng
claude plugin install typescript@outcomeeng
# Optional plugins
claude plugin install prose@outcomeeng
claude plugin install instructions@outcomeengAfter adding the marketplace, enable only the plugins a product needs in that repo's committed .codex/config.toml:
[plugins."product-tree@outcomeeng"]
enabled = true
[plugins."instructions@outcomeeng"]
enabled = trueAdd language plugins in projects that use them:
[plugins."python@outcomeeng"]
enabled = true
[plugins."typescript@outcomeeng"]
enabled = trueAdd domain plugins the same way:
[plugins."frontend@outcomeeng"]
enabled = true
[plugins."visual@outcomeeng"]
enabled = true
[plugins."hdl@outcomeeng"]
enabled = true> /bootstrap # set up a new Product Tree
> /author outcome for search # author a new outcome node
> /author PDR for auth policy # author a product decision
> /author ADR for caching strategy # author an architecture decision
> /apply # start the TDD flow
> /commit-changes # commit with Conventional Commits
See the full tutorial for the complete workflow — from bootstrapping to handoffs.
claude plugin marketplace update outcomeengcodex plugin marketplace upgrade outcomeengRepository maintainers can verify both catalogs with
just verify-marketplace-installation. The command installs each full catalog
and a generated valid subset through the real Claude Code and Codex CLIs in
disposable homes. The command also rejects an invalid subset that omits product-tree. It places
each installed plugin's generated Codex agent definitions in the disposable
home's agents/ directory and leaves persistent plugin state unchanged.
After merged distribution changes, just install-marketplace reads the
marketplace's name from the committed catalog and its source from the machine's
own registry, refreshes the registered clone, and brings every project- or
local-scope Claude Code install record of a cataloged plugin on the machine to
the head of that source's default branch: the invocation checkout's own records
through the native plugin update, every other such record by rewriting its
installed_plugins.json entry to the head's install path, version, and commit,
whether or not the record's directory still exists. No command runs with any
other checkout as its working directory, and no other checkout's settings or
project directory is inspected. A record of a plugin id the catalog renames
moves to the successor plugin's entry; any other record outside project and
local scope or outside the catalog is reported and left unchanged. The invocation checkout's
own settings matter only to bootstrap: settings it cannot read stop bootstrap
and nothing else. The run also refreshes the selected $CODEX_HOME, plans from
one listing read, reads the listing again after execution, reports each record's
version before and after, and exits nonzero when any project- or local-scope
record of a cataloged plugin is off the head afterwards.
The declared refresh selection uses Claude Code's install records
and Codex's raw selected-home plugin declarations, including disabled and uncached plugins.
Committed catalogs bound membership; trusted product configuration controls activation separately.
Implementation of this selection and activation-preserving recovery is tracked in
Change #48.
Empty state receives
only product-tree and a warning. The command rejects nonempty state without product-tree.
The command reconciles the selected plugins' generated Codex agent definitions
into the selected $CODEX_HOME/agents/ registry under a digest-bound marketplace
ownership record, leaving foreign and modified files untouched. A checkout
.codex/agents/ definition whose invoked skills live in the selected home is a
scope split that stops the run before mutation, per
spx/12-marketplace-state.adr.md; marketplace installation never requires
committing agent definitions into a checkout.
When working on the marketplace itself, every branch that changes a plugin's
distribution surface bumps that plugin's version exactly once. The just bump
recipe automates this — it detects which plugins changed under src/plugins/<name>/**
since origin/main, classifies each plugin's change pattern into a semver
segment, and updates the version field in every manifest each plugin owns
(.claude-plugin/plugin.json and, when present, .codex-plugin/plugin.json)
in lockstep:
just bump # auto-detect segment per plugin vs origin/main
just bump-dry # preview without writing
just bump-check # CI gate: exit non-zero if any changed plugin needs a bump
just bump origin/main minor # force minor for every changed plugin (warns on disagreements)
just bump main major # major bump vs local main (must be explicit)By default, the segment is detected per plugin from the file-status pattern of its changes:
| Plugin's changes include… | Segment |
|---|---|
Added / deleted / renamed skills/<slug>/SKILL.md |
minor |
Added / deleted / renamed agents/<slug>.md |
minor |
Added .claude-plugin/plugin.json or .codex-plugin/plugin.json (whole plugin or new surface) |
minor |
| Anything else — modifications to existing files, internal helpers, templates, references, fixtures | patch |
major is never auto-detected — it captures a deliberate stability commitment
and requires explicit --segment major. Passing an explicit --segment flag
overrides per-plugin detection and emits a stderr warning naming any plugin
whose detected segment differed (so you don't silently override the file-status
evidence). When a plugin already carries a bump on this branch, the bumper skips it and
bumps every other changed plugin in the same pass. It preserves manifest bytes
character-for-character outside the version field so bumps produce minimal diffs.
Only paths under src/plugins/<name>/** count as distribution-surface changes; edits
to spx/, AGENTS.md, tests, or other top-level files do not trigger a bump.
The installed plugin trees under dist/claude/ and dist/codex/ are generated from src/plugins/. After editing anything under src/plugins/, rebuild them:
just build-skills # uv run python -m outcomeeng.distribution.build src distThe pre-commit hook runs build-skills automatically, and just check-full's dist-diff step (git diff --exit-code dist) fails when dist/ is out of sync with src/, so each src/plugins/ change and its regenerated dist/ commit together. Never hand-edit dist/.
The spec-tree plugin is now product-tree. Five places in a consumer's setup can still name spec-tree, and each takes one step.
The marketplace catalog maps spec-tree to product-tree. After the marketplace refresh, the first Claude Code plugin command rewrites the spec-tree@outcomeeng key in .claude/settings.json to product-tree@outcomeeng, moves the key to the end of the enabledPlugins map, and drops spec-tree from claude plugin list. The rewrite changes a committed file, so commit it. product-tree stays enabled and not installed until you install it at the scope that holds the old record:
claude plugin marketplace update outcomeeng
claude plugin install product-tree@outcomeeng --scope projectOn Codex, refresh the marketplace, add the successor, and then remove the old plugin. After the refresh, spec-tree no longer appears in codex plugin list, and codex plugin remove spec-tree@outcomeeng still removes it, together with its entry in $CODEX_HOME/config.toml and its cache:
codex plugin marketplace upgrade outcomeeng
codex plugin add product-tree@outcomeeng
codex plugin remove spec-tree@outcomeengClaude Code rewrites the enabledPlugins key in .claude/settings.json as described above. No Codex command edits a project's committed .codex/config.toml: replace [plugins."spec-tree@outcomeeng"] with [plugins."product-tree@outcomeeng"] in that file by hand.
Run the update-instruction-block skill of the product-tree plugin in each repository. It replaces the SPEC-TREE router block and shared regions in CLAUDE.md and AGENTS.md with their PRODUCT-TREE forms once.
Codex reads subagent definitions from $CODEX_HOME/agents. Run the product-tree-plugin skill in Codex with the upgrade verb. It writes the product-tree_*.toml definitions and prunes the spec-tree_*.toml definitions that the marketplace ownership record lists, while their content still matches the record. A definition you changed, or replaced with a link, stays in place and the step reports it as retained: delete that file by hand.
Run the Codex steps in this order:
- Refresh the marketplace with
codex plugin marketplace upgrade outcomeeng. - Add the successor with
codex plugin add product-tree@outcomeeng. - Remove the old plugin with
codex plugin remove spec-tree@outcomeeng. - Replace
[plugins."spec-tree@outcomeeng"]with[plugins."product-tree@outcomeeng"]in each project's.codex/config.toml. If a project's.codex/config.tomlnames nospec-tree@outcomeengtable, skip this step for that project. - Start a new Codex session, so the session loads
product-tree. - Run the
product-tree-pluginskill with theupgradeverb. - If the skill reports a file as
retained, delete that file by hand. - Start a new Codex session, so the session discovers the
product-treedefinitions in$CODEX_HOME/agents.
Run the update-instruction-block skill in the session of step 6.
Skills are available in both Claude Code and Codex, with generated plugin surfaces carrying the thin agents their target supports. Every skill and thin agent across every plugin is listed in the auto-generated catalog below, sourced from .claude-plugin/marketplace.json and the YAML frontmatter of each plugin's SKILL.md and agents/*.md. Run just docs to regenerate after touching any of those files; just check-full enforces freshness in CI.
/bootstrap in action — interactive product interview and scaffold
Coding-agent environments, communication, and change orchestration: /operate-prowl, /operate-herdr, /operate-agent-mail, /message-agents, /recover-prowl-agents, /orchestrate-change
| Type | Name | Purpose |
|---|---|---|
| Skill | /control-positions |
Pacing token spend across positions hosted in Prowl panes, running one control pass, or installing, checking, stopping, or restarting the scheduled position-control job |
| Skill | /control-token-spend |
Measuring coding-session token consumption, explaining context and cache costs, detecting spending bursts, or managing local usage-report schedules |
| Skill | /direct-positions |
Directing positions across Products: resuming the directing session after a compaction, observing, waking, or steering a position's pane, or reading a position's mail |
| Skill | /message-agents |
Sending a message record to a registered agent-mail name with its one-line doorbell, discovering a Prowl coding-agent recipient, or sending facts, ownership proposals, state reports, authorizations, or acknowledgements to another agent pane |
| Skill | /operate-agent-mail |
A workflow registers a mail identity, sends a message record, lists a recipient's records, reads or acknowledges one message, or waits for a later record in the agent-mail store |
| Skill | /operate-herdr |
Running a public herdr operation — agent inventory, read, bounded wait, prompt, start, relaunch, stop, keystroke, worktree create, or worktree open — on agent sessions herdr hosts |
| Skill | /operate-prowl |
A workflow needs a public Prowl operation or a correlated delegation handback between Prowl coding agents |
| Skill | /orchestrate-change |
Starting the Applier session for a Change, or checking, restarting, or answering a running Applier |
| Skill | /recover-prowl-agents |
Preparing for or recovering coding-agent sessions after a Prowl restart |
| Skill | /wait-for-event |
A coding-agent workflow must wake on the first of an agent-mail record, a herdr-hosted agent session state change, a consumption-control alert, or a clock instant, within one bounded timeout |
Contributions to repositories you do not control: /upstream, /open-upstream-pr, /manage-upstream-pr, /open-upstream-issue, /manage-upstream-issue, /sync-fork
| Type | Name | Purpose |
|---|---|---|
| Skill | /contribution-standards |
Contribution standards for changes and reports sent to a repository the operator does not control |
| Skill | /manage-upstream-issue |
Continuing an open issue in a repository the operator does not control — answering a maintainer, adding evidence, or reporting the thread's current state |
| Skill | /manage-upstream-pr |
Continuing an open pull request against a repository the operator does not control — answering review, publishing a revision, or reporting its current state |
| Skill | /open-upstream-issue |
Filing an issue in a repository the operator does not control — a fork's upstream, or any base whose permission is READ, TRIAGE, or NONE |
| Skill | /open-upstream-pr |
Opening a pull request against a repository the operator does not control — a fork's upstream, or any base whose permission is READ, TRIAGE, or NONE |
| Skill | /sync-fork |
Bringing a fork's default branch current with the repository it was forked from |
| Skill | /upstream |
Starting a contribution to a repository the operator does not control, or when a contribution step needs that target and no live UPSTREAM_TARGET marker carries it — it resolves the upstream a fork came from, the head to push from, and the operator's permission on it |
Frontend design: /design-frontend skill
| Type | Name | Purpose |
|---|---|---|
| Skill | /design-frontend |
Designing or building web components, pages, or dashboards |
Go engineering: /author-go-test, /author-go-code, /audit-go-code, /audit-go-test, /audit-go-architecture, /author-go-architecture, /simplify-go-code
| Type | Name | Purpose |
|---|---|---|
| Skill | /audit-go-architecture |
Go-specific architecture audit |
| Skill | /audit-go-code |
Go implementation-code audit methodology |
| Skill | /audit-go-test |
Go test-evidence audit methodology |
| Skill | /author-go-architecture |
Writing ADRs for Go. NEVER author a Go ADR without this skill |
| Skill | /author-go-code |
Writing or fixing implementation code for Go |
| Skill | /author-go-test |
Writing or fixing tests for Go |
| Skill | /go-architecture-standards |
Go ADR conventions enforced across architect and auditor skills |
| Skill | /go-code-standards |
Go code standards enforced across all skills |
| Skill | /go-test-standards |
Go test standards enforced across all skills |
| Skill | /simplify-go-code |
Simplifying Go implementation while preserving behavior |
HDL engineering: /review-vhdl, /review-systemverilog (idiomatic VHDL-2008 and SystemVerilog IEEE 1800-2017 review)
| Type | Name | Purpose |
|---|---|---|
| Skill | /review-systemverilog |
Reviewing SystemVerilog or Verilog code for idiomatic style, synthesizability, or best practices |
| Skill | /review-vhdl |
Reviewing VHDL code for idiomatic style, synthesizability, or best practices |
Instruction authoring: /author-skill, /author-subagent
| Type | Name | Purpose |
|---|---|---|
| Skill | /agent-prompt-standards |
Agent prompt writing conventions enforced across all creator and auditor skills |
| Skill | /audit-skill |
SKILL.md audit methodology |
| Skill | /audit-subagent |
Claude: Subagent-configuration audit methodology; Codex: Custom agent-configuration audit methodology |
| Skill | /author-skill |
Creating, editing, or improving SKILL.md files or bundled workflows, references, templates, and scripts |
| Skill | /author-subagent |
Claude: Creating, editing, or configuring subagents; Codex: Creating, editing, or configuring custom agents |
| Skill | /skill-standards |
Skill authoring standards enforced across all creating and auditing skills |
| Skill | /subagent-standards |
Claude: Configuration, configuration-subject, profile, capability, invocation, placement, and evidence standards for subagents, with their rule catalog; Codex: Configuration, configuration-subject, profile, capability, invocation, placement, and evidence standards for custom agents, with their rule catalog |
| Agent | skill-auditor |
Auditing, reviewing, or evaluating SKILL.md files for best practices compliance, or when the user asks to audit a skill |
| Agent | subagent-auditor |
Claude: Auditing, reviewing, or evaluating subagent configuration files for best practices compliance, or when the user asks to audit a subagent; Codex: Auditing, reviewing, or evaluating custom agent configuration files for best practices compliance, or when the user asks to audit a custom agent |
Product Tree: /understand, /contextualize, /bootstrap, /author, /decompose, /refactor, /align, /interview, /test, /audit-tests, /audit-eval-evidence, /audit-adr, /audit-pdr, /audit-implementation, /audit-specs, /audit-changeset-coherence, /apply, /commit-changes, /manage-change, /refocus, /diagnose
| Type | Name | Purpose |
|---|---|---|
| Skill | /align |
Reviewing, auditing, or checking spec file conformance |
| Skill | /apply |
Implementing any product-tree work item |
| Skill | /apply-change |
A session applies one claimed Applicable Change as its Applier |
| Skill | /audit-adr |
ADR audit methodology |
| Skill | /audit-assertions |
Audit-assertion methodology |
| Skill | /audit-change |
Change record audit methodology |
| Skill | /audit-changeset-coherence |
Changeset-coherence audit methodology |
| Skill | /audit-eval-evidence |
Eval-evidence audit methodology |
| Skill | /audit-implementation |
Implementation audit methodology |
| Skill | /audit-pdr |
PDR audit methodology |
| Skill | /audit-specs |
Spec-node audit methodology |
| Skill | /audit-tests |
Test-evidence audit methodology |
| Skill | /author |
Adding, defining, or creating specs, decisions, or nodes |
| Skill | /author-change |
Recording, refining, publishing, splitting, merging, or reframing an Outcome Engineering Change record, or when turning a request into one |
| Skill | /bootstrap |
Setting up a new Product Tree in an empty or absent spx/ directory |
| Skill | /change-standards |
Change record standards for metadata, body, Declarations, Maturity, Transitions, structure, the form check, and the GitHub Change Store |
| Skill | /commit-changes |
Committing changes or when user says "commit" |
| Skill | /contextualize |
Asking about status, progress, or what exists in the Product Tree |
| Skill | /decompose |
Breaking down, splitting, scoping, composing, or structuring Product Tree nodes |
| Skill | /diagnose |
Diagnosing the health of a product-tree or spx environment, when checking whether the SessionStart hook fired for the current session, or when troubleshooting a missing session identity, worktree claim, or unreachable spx CLI |
| Skill | /implement-change |
Language-implementation workflow for one product-tree node: the target language's installed architect, code, and simplify skills run in one session, invoked with a canonical node path |
| Skill | /init-worktrees |
Setting up a repository's git worktree layout — classifying a checkout as a single tree, a bare-repo worktree pool, or non-compliant, and provisioning the bare-repo pool while pushing every local ref to the remote and carrying a prior checkout's gitignored state across |
| Skill | /inspect-github-actions |
The user asks about CI failures, workflow logs, GitHub Actions status, pipeline issues, or troubleshooting failed builds |
| Skill | /interview |
Requirements interviews, draft approval, or unresolved scope/design questions while creating or modifying an artifact (spec, ADR, PDR, test, code, doc) |
| Skill | /manage-change |
Claiming, releasing, submitting, advancing, lowering, closing Applied, or closing Abandoned a Change in the declared Change Store |
| Skill | /manage-github-pr |
The user asks to open or manage a GitHub pull request |
| Skill | /manage-pr |
Composed re-entry protocol for an already open PR: foreground check and review waits, re-entry after compaction or push, the merge-readiness decision, and merge command selection |
| Skill | /merge |
The user asks to ship, integrate, or merge a changeset into the default branch on origin, or runs /merge |
| Skill | /merging-standards |
Shared merge-lifecycle invariants and routing for detailed preflight, branch, review, authority-gate, delivery, and closeout policy |
| Skill | /migrate-3-to-4 |
Converting the decision citations of a 3.x Product Tree product to the tree-absolute Markdown links the 4.0 methodology requires, or when spx rejects a decision cited as a code span, a bare path, or a node-local link |
| Skill | /open-pr |
PR opening protocol for VERIFICATION_READINESS, branch push, ready PR creation, and the first management pass |
| Skill | /project-run-journal |
Verification run-journal protocol |
| Skill | /refactor |
Moving nodes, re-scoping content, or extracting shared providers |
| Skill | /refocus |
Running ad hoc commands or writing code without a spec inside a repository whose root contains spx/ |
| Skill | /report-changes |
Listing, reporting, or checking the open Changes in the Change store, such as which Changes are waiting, blocked, or claimed |
| Skill | /review-changes |
Reviewing working changes on a branch against a base ref |
| Skill | /scope-changeset |
Committed changeset endpoint identities and changed paths, resolved through the canonical Git scope capability |
| Skill | /select-artifacts |
The registered artifacts and audit skills each changed path selects, resolved through the artifact registry the build renders beside this skill's reader |
| Skill | /slice |
Selecting the next executable slice to implement or deciding which product-tree nodes to build next from an implementation plan |
| Skill | /sync-base |
ALWAYS invoke this skill to bring a branch behind its base current — before reading product truth, before verifying, and before every merge push |
| Skill | /test |
Writing or repairing deterministic tests for a spec assertion, selecting a decision Testing rule's assertion type, or when learning the testing approach |
| Skill | /test-evidence-standards |
Shared test-evidence standards — seam, case provenance, oracle independence, execution level, per-assertion-type artifact permissions, and the pre-authoring assertion-design record |
| Skill | /understand |
The live PRODUCT_TREE_FOUNDATION marker is absent before direct filesystem access under spx/ or before reading, searching, listing, or changing source or test files |
| Skill | /update-instruction-block |
Manually regenerating, refreshing, or scaffolding the managed Product Tree instruction surface of a product's two root instruction files, AGENTS.md and its sibling, from the installed product-tree template, or reconciling a shared region that differs between the two files |
| Skill | /verify |
Selecting or establishing evidence for spec assertions, decision verification rules, or a product-tree scope |
| Skill | /wait-for-load |
Starting a resource-intensive local command or when host load is high |
| Agent | adr-auditor |
Verifier running /audit-adr: a calling skill launches it by exact name with one ADR path when its gate requires an independent audit of that ADR |
| Agent | assertion-auditor |
Verifier running /audit-assertions: a calling skill launches it by exact name with one node spec path when its gate requires an independent verdict on that node's Audit Assertions |
| Agent | change-applier |
ALWAYS select at session start when an agent session applies one claimed Applicable Change as its Applier through /apply-change |
| Agent | change-auditor |
Audits one stored Outcome Engineering Change Leaf in an isolated verifier context |
| Agent | change-author |
Round session /apply-change launches by exact name when a Declarations subsection needs product-tree artifacts: decisions, specs, nodes, and notes produced or repaired through /author |
| Agent | change-implementer |
Round session /apply-change launches by exact name when a Declarations subsection needs a language implementation of one node produced or repaired through /implement-change |
| Agent | change-skill-author |
Round session /apply-change launches by exact name when a Declarations subsection needs a skill surface (a SKILL.md, another file in a skill directory, or an authored shared fragment) produced or repaired through instructions:author-skill |
| Agent | change-subagent-author |
Round session /apply-change launches by exact name when a Declarations subsection needs a subagent definition produced or repaired through instructions:author-subagent |
| Agent | change-tester |
Round session /apply-change launches by exact name when a Declarations subsection needs routed [test] evidence produced or repaired through /test |
| Agent | change-verifier |
Round session /apply-change launches by exact name when a Declarations subsection needs assertions routed to their verification type and evidence produced or repaired through /verify |
| Agent | changes-reviewer |
Reviews working changes against a base ref in an isolated verifier context |
| Agent | changeset-coherence-auditor |
Audits whether one exact committed changeset is one coherent review unit or requires a dependency-ordered split, in an isolated read-only verifier context |
| Agent | eval-evidence-auditor |
Verifier running /audit-eval-evidence: a calling skill launches it by exact name with one spec node path when its gate requires an independent audit of that node's eval evidence |
| Agent | implementation-auditor |
Verifier running /audit-implementation: a calling skill launches it by exact name with one changeset scope selector when its gate requires an independent audit of that changeset's implementation |
| Agent | instruction-block-updater |
Updates the managed Product Tree instruction block in a product's root CLAUDE.md and AGENTS.md without operator interaction, in an isolated context |
| Agent | pdr-auditor |
Verifier running /audit-pdr: a calling skill launches it by exact name with one PDR path when its gate requires an independent audit of that PDR |
| Agent | spec-auditor |
Verifier running /audit-specs: a calling skill launches it by exact name with one node spec path when its gate requires an independent audit of that spec |
| Agent | test-evidence-auditor |
Verifier running /audit-tests: a calling skill launches it by exact name with one spec node path when its gate requires an independent audit of that node's test evidence |
Prose craft through a matched triple (/author-prose-architecture, /author-prose, /audit-prose) over caller-supplied kind layers: interface, documentation, copy
| Type | Name | Purpose |
|---|---|---|
| Skill | /audit-prose |
Prose audit methodology |
| Skill | /author-prose |
Writing text for human readers — docs pages, UI text, error messages, emails, release notes |
| Skill | /author-prose-architecture |
Writing ADRs for prose |
| Skill | /prose-architecture-standards |
Prose ADR conventions enforced across architect and auditor skills |
| Skill | /prose-standards |
Prose standards enforced across all skills |
| Agent | prose-auditor |
Auditing human-facing text — documents, web pages, articles, docs, UI text, product messages, or internal team pages — for prose quality and style-kind conformance |
Python engineering: /author-python-test, /author-python-code, /audit-python-code, /audit-python-test, /audit-python-architecture, /author-python-architecture
| Type | Name | Purpose |
|---|---|---|
| Skill | /audit-python-architecture |
Python-specific architecture audit |
| Skill | /audit-python-code |
Python implementation-code audit methodology |
| Skill | /audit-python-test |
Python test-evidence audit methodology |
| Skill | /author-python-architecture |
Writing ADRs for Python |
| Skill | /author-python-code |
Writing or fixing implementation code for Python |
| Skill | /author-python-test |
Writing or fixing tests for Python |
| Skill | /python-architecture-standards |
Python ADR conventions enforced across architect and auditor skills |
| Skill | /python-code-standards |
Python code standards enforced across all skills |
| Skill | /python-test-standards |
Python test standards enforced across all skills |
Rust engineering: /author-rust-test, /author-rust-code, /audit-rust-code, /audit-rust-test, /audit-rust-architecture, /author-rust-architecture, /simplify-rust-code
| Type | Name | Purpose |
|---|---|---|
| Skill | /audit-rust-architecture |
Rust-specific architecture audit |
| Skill | /audit-rust-code |
Rust implementation-code audit methodology |
| Skill | /audit-rust-test |
Rust test-evidence audit methodology |
| Skill | /author-rust-architecture |
Writing ADRs for Rust |
| Skill | /author-rust-code |
Writing or fixing implementation code for Rust |
| Skill | /author-rust-test |
Writing or fixing tests for Rust |
| Skill | /rust-architecture-standards |
Rust ADR conventions enforced across architect and auditor skills |
| Skill | /rust-code-standards |
Rust code standards enforced across all skills |
| Skill | /rust-test-standards |
Rust test standards enforced across all skills |
| Skill | /simplify-rust-code |
Simplifying Rust implementation while preserving behavior |
TypeScript engineering: /author-typescript-test, /author-typescript-code, /audit-typescript-code, /audit-typescript-test, /audit-typescript-architecture, /author-typescript-architecture, /simplify-typescript-code
| Type | Name | Purpose |
|---|---|---|
| Skill | /audit-typescript-architecture |
TypeScript-specific architecture audit |
| Skill | /audit-typescript-code |
TypeScript implementation-code audit methodology |
| Skill | /audit-typescript-test |
TypeScript test-evidence audit methodology |
| Skill | /author-typescript-architecture |
Writing ADRs for TypeScript |
| Skill | /author-typescript-code |
Writing or fixing implementation code for TypeScript |
| Skill | /author-typescript-test |
Writing or fixing tests for TypeScript |
| Skill | /simplify-typescript-code |
Simplifying TypeScript implementation while preserving behavior |
| Skill | /typescript-architecture-standards |
TypeScript ADR conventions enforced across architect and auditor skills |
| Skill | /typescript-code-standards |
TypeScript code standards enforced across all skills |
| Skill | /typescript-test-standards |
TypeScript test standards enforced across all skills |
Work deliverables: /draw-excalidraw (Excalidraw diagrams), /sanitize-powerpoint (PowerPoint deck cleanup)
| Type | Name | Purpose |
|---|---|---|
| Skill | /draw-excalidraw |
Creating Excalidraw diagrams, visualizing workflows, architectures, or concepts |
| Skill | /sanitize-powerpoint |
Sanitizing, cleaning up, auditing, or aligning a PowerPoint (.pptx) deck — slide-master and layout structure, layout type attributes, stray fonts, non-theme colors, or layout naming |
Skills are distributed as standalone repositories, compatible with any agent that supports the Agent Skills open standard.
| Repository | Purpose | Install |
|---|---|---|
| spec-tree | Product Tree methodology skills for Outcome Engineering | npx skills add outcomeeng/spec-tree |
| python | Python engineering skills | npx skills add outcomeeng/python |
| typescript | TypeScript engineering skills | npx skills add outcomeeng/typescript |
| foundation | Foundation skills (prose, plugin development, frontend) | npx skills add outcomeeng/foundation |
- Codex
- Codex Overview
codex plugin --helpcodex plugin marketplace --help
The instructions plugin's meta-skills are derived from TÂCHES Claude Code Resources. The claim and release Transitions of /manage-change descend from the former /handoff and /pickup commands, which were based on /whats-next from the same project.
MIT





